Can Stupid AI Make Better Creative Work?
Useful creative ideas do not always look sensible at first. At BRXND NYC 2026, Springboards' Pip Bingemann argues for AI that produces a wider range of starting points. Through live demonstrations of Flint and an experimental Dada model, he explores creative sameness and the role of people in deciding which surprising directions are worth developing.
Key takeaways
- Watch for convergence across AI suggestions, including outputs that look individually plausible.
- Use divergent responses to open up creative territory; people still need to judge and develop the ideas.
- Leave space for playful experiments that challenge the most predictable answer.
In this session
00:00 Introducing artificial stupidity
Bingemann introduces Springboards and his deliberately provocative case for artificial stupidity.
00:58 The artificial hivemind and creative sameness
He discusses the Artificial Hivemind paper and examples of models returning similar creative suggestions.
04:29 Why distinctive ideas can look wrong at first
Examples from sport, culture and advertising support his argument that distinctive ideas can initially look strange or wrong.
05:26 Flint live demo: numbers, cars and taglines
A live comparison asks models for numbers, cars and shoe taglines to illustrate the kinds of divergence Flint is designed to produce.
08:25 Why turning up temperature is not the whole answer
Bingemann uses a joke-generation demo to explain why he does not consider a higher temperature setting a sufficient creative solution.
09:24 Critical tokens and branching into different ideas
He describes identifying branching points in generation where greater variation can send an idea down a different path.
11:04 The experimental Dada model
The experimental Dada model produces unusual dog names and responses about the meaning of life, illustrating a more playful direction.
13:32 Creative ambition and the case for being stupid
He closes with a call for creative ambition and a willingness to explore ideas that do not immediately look smart.
Recorded at BRXND NYC on November 5, 2026, at The TimesCenter in New York.
Written by
.png&w=3840&q=75)
Alephic Writing
Ideas for building the AI-native enterprise.
New essays on AI, marketing, software, and the organizations being rebuilt around them.
